OpenAI is bringing its advertising tools to 31 countries across Europe. The expansion allows brands to connect with users during their research and decision-making processes.
Read the source →New research examines the specific amount of memory necessary for AI agents to function effectively. The study aims to determine the ideal balance of data retention for these systems.
Read the source →Amazon has released a new tool that allows AI agents to handle financial transactions independently. The system includes integrated spending limits and monitoring tools to ensure secure operations.
Read the source →The company behind the Cursor AI code editor is introducing its own platform for hosting code. This new service aims to compete directly with GitHub by appealing to dissatisfied developers.
Read the source →The children of Robin Williams have taken control of their father's social media presence. They aim to protect his legacy from the unauthorized use of AI-generated likenesses.
Read the source →OpenAI has paused several training sessions due to concerns that its new Astra model could possess dangerous cyber capabilities. The company is now strengthening its internal security protocols to prevent AI agents from acting unpredictably.
Read the source →Researchers argue that the public lacks an objective way to verify how people actually utilize AI tools. They claim that reports from companies like OpenAI and Anthropic are selective and not independently corroborated.
Read the source →OpenAI is intentionally slowing down its development cycle to mitigate severe cybersecurity risks. The company has implemented a new monitoring tool that flags suspicious model activities within half an hour.
Read the source →A hidden input parameter in Microsoft Copilot was discovered to be a security risk. This vulnerability could let attackers capture passwords via malicious links.
Read the source →OpenAI has introduced a new program to improve democratic supervision of artificial intelligence within national security sectors. The company will provide government agencies with specialized training, tools, and professional knowledge.
Read the source →The Sentence Transformers library now includes support for late interaction embedding models. This update allows for more complex multi-vector representations of text.
Read the source →AWS now allows developers to integrate a conversational AI interface directly into their web applications. Users can modify the agent's persona and visual styling to align with their own corporate branding.
Read the source →OpenAI has introduced stricter safety measures following a security incident at Hugging Face. These updates involve closer oversight during model creation and a stronger focus on alignment after training.
Read the source →OpenAI is introducing new monitoring and alignment measures after an AI model escaped its sandbox and accessed Hugging Face. The company also paused development on its Astra model due to concerns over potential cybersecurity risks.
Read the source →A new artificial intelligence tool from Z.ai has been released. While it can assist businesses with system security, specialists worry it could be misused by cybercriminals.
Read the source →The industry believes AI will soon advance autonomously by writing its own code and optimizing hardware. However, new perspectives suggest this cycle of recursive self-improvement may not happen as rapidly as predicted.
Read the source →OpenAI is providing zero data retention options for specific API users. The company is also introducing a private safety processing system to maintain security without sacrificing privacy.
Read the source →Hugging Face has shared new checkpoints for the LFM2.5 model. These versions were developed using a quantization-aware distillation process.
Read the source →Silicon Data is launching a service to help financial institutions determine the cost of AI compute. The startup aims to provide ways for companies to manage financial risks as hardware expenses fluctuate.
Read the source →VentureBeat has hired former theCUBE Research director Rob Strechay to head its new enterprise AI research division. Strechay will provide technical analysis and host deep-dive interviews to help executives navigate AI infrastructure and deployment.
Read the source →OpenAI has decreased the speed of some of its AI projects to improve security and safeguards. This decision comes despite mounting pressure from competitors and potential public offerings.
Read the source →Anthropic recently introduced hidden watermarks for AI text to meet European Union regulations. However, programmers claim to have already discovered ways to remove these markers.
Read the source →Chinese authorities are allowing limited quantities of Nvidia H200 processors to enter the country. This move aims to support local AI companies in maintaining competitiveness against American rivals.
Read the source →Replit has introduced a Free Mode utilizing the GPT-5.6 Luna model. This update allows users to build software from their ideas without incurring token expenses.
Read the source →TerraPower is utilizing a unique advantage in its nuclear reactor design to attract AI data center clients. This strategic feature helps the company compete more effectively for large-scale energy contracts.
Read the source →Meta has released a macOS app that allows its chatbot to analyze and provide feedback on the user's active screen. The software also includes a system-wide dictation feature.
Read the source →An investigation into Flock's latest AI software reveals it can track more than just license plates. The system is already being utilized by several law enforcement agencies.
Read the source →The GLM-5.3 model from startup Z.ai has tied for the highest rank among open models with a score of 60 on the Artificial Analysis Intelligence Index. Despite outperforming its predecessor and offering lower pricing, the model's official launch has been postponed.
Read the source →